It was the recent articles in audioXpress ( March/April 08 ) and your response in the recent issue that prompted me to explore diffraction.
In your response you stated " Diffraction - that aspect of horns that horn theory cannot deal with - turns out to be of predominate importance to sound quality...a very bad thing...Any form of discontinuity anywhere in the device can produce this problem"
So far I have found little to expand my understanding on the subject in the context of acoustics.
I am familar with the diffraction in optics, but didn't know how it would apply, and when it falls apart.
Most analysis that I've found is limited to optics and the "classic" experiments of Huygens.
So far the consensus is as Linkwitz said " Diffraction is a difficult subject."
There are a number of questions that I have about diffraction, raised by these articles, where/what would be the appropriate manner to ask them?
Thanks for the response
In your response you stated " Diffraction - that aspect of horns that horn theory cannot deal with - turns out to be of predominate importance to sound quality...a very bad thing...Any form of discontinuity anywhere in the device can produce this problem"
So far I have found little to expand my understanding on the subject in the context of acoustics.
I am familar with the diffraction in optics, but didn't know how it would apply, and when it falls apart.
Most analysis that I've found is limited to optics and the "classic" experiments of Huygens.
So far the consensus is as Linkwitz said " Diffraction is a difficult subject."
There are a number of questions that I have about diffraction, raised by these articles, where/what would be the appropriate manner to ask them?
Thanks for the response
Well this forum is not bad, but this topic is maybe not ideal.
One of the problems with "diffraction" is that it can come in many forms. The most common is what happens when a wave encounters a "knife edge" - the classic optics problem. We know that the field that is created can be thought of as an incident field plus a diffractioon field which is approximated as a wave that originates at the knife edge. This is pretty close to being correct but still not exact. But the point of significance is that virtually all diffraction problems can be approximated as the original source plus a new source that eminates from the diffraction point. Its easy to see that when these two waves arrive at the listener they have traveled different paths of different lengths. This is where the problems come from. The ear apears to be quite sensitive to these different path lengths, i.e. arrival times. It appears that the ear masks very well in the frequency domain but not very well at all in the time domain. This makes the time delayed diffracted signal very audible despite its usually low level.
Any discontinuity in the boundaries of a horn will cause diffraction. And I mean any. Even a slowly changing boundary will diffract - obviously sharp ones do. Thus it seems pertinent to look for boundary shapes that diffract the least. This can be done and in fact the OS waveguide is just that boundary. No shape connnecting the same two aperatures can diffract less.
Optics deals with wavelengths that are almost always short compared to the objects (a lense is many many wavelengths in size), while acoustics will have wavelengths shorter, equal and longer - the whole gamut. Diffraction changes quite a bit with these different regions which makes it more complex in acoustics than it is in optics.
I studied optics in school to study diffraction. It was instructive, but as I said, the wavelngth domains where quite different.
One of the problems with "diffraction" is that it can come in many forms. The most common is what happens when a wave encounters a "knife edge" - the classic optics problem. We know that the field that is created can be thought of as an incident field plus a diffractioon field which is approximated as a wave that originates at the knife edge. This is pretty close to being correct but still not exact. But the point of significance is that virtually all diffraction problems can be approximated as the original source plus a new source that eminates from the diffraction point. Its easy to see that when these two waves arrive at the listener they have traveled different paths of different lengths. This is where the problems come from. The ear apears to be quite sensitive to these different path lengths, i.e. arrival times. It appears that the ear masks very well in the frequency domain but not very well at all in the time domain. This makes the time delayed diffracted signal very audible despite its usually low level.
Any discontinuity in the boundaries of a horn will cause diffraction. And I mean any. Even a slowly changing boundary will diffract - obviously sharp ones do. Thus it seems pertinent to look for boundary shapes that diffract the least. This can be done and in fact the OS waveguide is just that boundary. No shape connnecting the same two aperatures can diffract less.
Optics deals with wavelengths that are almost always short compared to the objects (a lense is many many wavelengths in size), while acoustics will have wavelengths shorter, equal and longer - the whole gamut. Diffraction changes quite a bit with these different regions which makes it more complex in acoustics than it is in optics.
I studied optics in school to study diffraction. It was instructive, but as I said, the wavelngth domains where quite different.
This is really interesting - I have a (very) basic understanding of diffraction, but never considered it in terms of GD.
A couple of questions:
1) I'm working on some ideas to create spherical speaker enclosures, so the effective radius of the front baffle would be, say, 10cm (0.1m). If my understanding is correct, this means it should reduce diffraction from:
(343m/s / 0.1m) / 8 = 428.75Hz
(or 1715Hz for where the effect is halved)
Would that be considered sufficient to have a major effect at eliminating the GD from diffractions? I take it that the diffraction based GD is considered more damaging to sound quality than that which it's usually masking (i.e. the end result would be a positive step)?
2) I'm familiar with GD being caused by vents/ports, but what would normally cause GD at higher frequencies (such as 2kHz)?
A couple of questions:
1) I'm working on some ideas to create spherical speaker enclosures, so the effective radius of the front baffle would be, say, 10cm (0.1m). If my understanding is correct, this means it should reduce diffraction from:
(343m/s / 0.1m) / 8 = 428.75Hz
(or 1715Hz for where the effect is halved)
Would that be considered sufficient to have a major effect at eliminating the GD from diffractions? I take it that the diffraction based GD is considered more damaging to sound quality than that which it's usually masking (i.e. the end result would be a positive step)?
2) I'm familiar with GD being caused by vents/ports, but what would normally cause GD at higher frequencies (such as 2kHz)?
Yes, a spherical enclosure will have the lowest possible diffraction of any enclosure shape, but it is a little tough to make and not really required. A cabinet with generously radiused edges seems to do the trick, but I mean big radi on the edge, like 5 cm radi.
At high freqencies GD would almost exclusively be from the crossover, if not done correctly or diffraction. In a well designed system crossover the system will be limited in GD by the enclosure diffraction. There are some things in a driver that can cause GD and there is a lot of diffraction in a horn if you arn't careful, but with those things under control, the cabinet remains as the worst culprit.
At high freqencies GD would almost exclusively be from the crossover, if not done correctly or diffraction. In a well designed system crossover the system will be limited in GD by the enclosure diffraction. There are some things in a driver that can cause GD and there is a lot of diffraction in a horn if you arn't careful, but with those things under control, the cabinet remains as the worst culprit.
nigelwright7557 said:
Doppler type intermodulation distortion and beaming become a problem with one full range speaker. A wide bandwidth midrange makes sense - say 100-1.5Khz - but a 3 way speaker with Xovers seems necessary for the best sound. A steep xover slope on the woofer-midrange can put a wider frequency range on your 1 extended range speaker, but with the phase shift issues being discussed in this thread.
How do you get the most bandwidth on one speaker with low IM distortion and low group delay? Digital xovers may be the best solution.
sploo:
When I first got interested in spherical and other ovoid enclosures there were few construction options available for DIY.
Today there are more options including pre-made
http://www.usenclosure.com/
I made 8" spheres very inexpensively, but it's labor intensive, and to my ears well worth the effort, much better than a box with a small radius round-over.
Mr Geddes:
I also studied optics early, and waveguides in the context of microwave communications during my electronics education. There was no acoustics degree offered.
Since light and radio waves are not sound I did not want to wrongly apply their analysis and characteristics.
I have yet to find much on this ( diffraction ) in acoustics. It also appears that there is little or no software to simulate diffraction; other that the simple razor edge demos.
Because as you said it comes in many forms, perhaps I can submit 2 scenarios for you consideration:
1: In the context of folded Bass horns built for high SPL PA application: They obviously have many disruptions and changes in direction. Advocates of these horns assert that the bends and folds acts as low pass filters, and that because these horns are filtered electronically @ 24 - 48 db/ octave that issues like HOM are irrelevant. Opinion?
2: There is a design for a curved array that is unusual in that it sandwiched between 2 flat 1" thick rectangular sides. The center drivers are the furthest back from front edges, the drivers at both ends of the array are very close to the leading edges, with varying distances for the remaining drivers.I have no simulation for this, but I would suspect that having leading edges in front of an array is not a good idea for several reasons including diffraction off the sharp 90 degee edges. Opinion?
When I first got interested in spherical and other ovoid enclosures there were few construction options available for DIY.
Today there are more options including pre-made
http://www.usenclosure.com/
I made 8" spheres very inexpensively, but it's labor intensive, and to my ears well worth the effort, much better than a box with a small radius round-over.
Mr Geddes:
I also studied optics early, and waveguides in the context of microwave communications during my electronics education. There was no acoustics degree offered.
Since light and radio waves are not sound I did not want to wrongly apply their analysis and characteristics.
I have yet to find much on this ( diffraction ) in acoustics. It also appears that there is little or no software to simulate diffraction; other that the simple razor edge demos.
Because as you said it comes in many forms, perhaps I can submit 2 scenarios for you consideration:
1: In the context of folded Bass horns built for high SPL PA application: They obviously have many disruptions and changes in direction. Advocates of these horns assert that the bends and folds acts as low pass filters, and that because these horns are filtered electronically @ 24 - 48 db/ octave that issues like HOM are irrelevant. Opinion?
2: There is a design for a curved array that is unusual in that it sandwiched between 2 flat 1" thick rectangular sides. The center drivers are the furthest back from front edges, the drivers at both ends of the array are very close to the leading edges, with varying distances for the remaining drivers.I have no simulation for this, but I would suspect that having leading edges in front of an array is not a good idea for several reasons including diffraction off the sharp 90 degee edges. Opinion?
gedlee said:
Earl, don't say it's so
Seriously, reduction of diffraction was only one of a number of goals of my quest to build spherical enclosures, though it's useful to actually get some quantitative data on what benefit a level of radius might achieve.
Planet10 - many thanks for the link. It's something I saw a few years back, and was one of my inspirations for going down the spherical route.
gedlee said:
Ah yes, forgot about crossover issues (don't think I'd quite engaged brain when I wrote the original question). It's good news from my point of view if the majority of effect is from the cabinet, as it's something I'm obviously looking at combating.
Staying on topic - I was party to an ABX test a few years back, using audio files with simulated group delay (in the low frequencies), and the results were pretty clear - out of 10 or so participants, and several tens of trials, even pretty large values (tens of ms) were considered inaudible. It's quite a kick to your ego when you can't distinguish something you reckon you should be able to hear!
HK26147 said:
Interesting - I wasn't aware there were places you could buy ready made enclosures of this type. I wonder what this magical material is that they claim to be making their boxes from?
How did you construct your spheres? My route is one of CNC and hand labour, mixed with some software I designed.
My program takes the details of a driver, the desired enclosure volume, and a specified cross section (curve) and attempts to construct an object of sufficient size to take the driver, satisfy the volume requirements (taking into account wall thickness and bracing) and works out the best way to make it from sheets of a given thickness (e.g. 3/4").
It then dumps out circles (for each layer) into AutoCAD, from which they are CNC cut, then glued together, then lathed to produce the desired outer shape. A bit time consuming, hence I'm looking to produce them using a casting process. Incidentally, Earl has also contributed to that with his casting knowledge, on another thread.
sploo
Right off the top, your tools and techniques are much more SOTA than mine; ( I don't have access to a CNC machine. )
I got interested in them ( non-boxes ) after articles in SB in the 80's mentioned the work of Olson ( as posted above ) and individuals began fabricating ovids. I bought a pair of Focal Eggs in the late 80's.
These were made of polyester resin/fiber.
Does a thread already exist for this sort of thing? If not, please start one detailing your works, and I'll include my humble efforts & research.
Right off the top, your tools and techniques are much more SOTA than mine; ( I don't have access to a CNC machine. )
I got interested in them ( non-boxes ) after articles in SB in the 80's mentioned the work of Olson ( as posted above ) and individuals began fabricating ovids. I bought a pair of Focal Eggs in the late 80's.
These were made of polyester resin/fiber.
Does a thread already exist for this sort of thing? If not, please start one detailing your works, and I'll include my humble efforts & research.
As far as I know US Enclosures uses a chop gun fiberglass system. It is inherently "multimodulus" because it isn't homogeneous, perhaps they add some secret ingredients, like sand or something, in layers, or mix fillers into their resin.
One of the principals (Brian O'Neill, IIRC) went around on a bunch of audio websites a few years ago and claimed he had sole rights to make spherical and egg shaped enclosures, and took a verbal beating everywhere because of it. He kept it up for a long time, then the posts stopped and a few years later their website switched from an AOL personal page under the name of kmetaverso to a dot com site and started touting large customers in its unique rather semi-literate fashion.
I recall him asking (on the Basslist or DIYloudspeakers list) for ideas for a business plan so they could get an SBA loan. I told him to market their speakers to little boutique mall shop chains like the Gap, etc as "Sound Spheres" that hang from the ceiling - a little searching might find that post. It appears they tout that sort of market as one of their customers on their website. Maybe I made them some money...
One of the principals (Brian O'Neill, IIRC) went around on a bunch of audio websites a few years ago and claimed he had sole rights to make spherical and egg shaped enclosures, and took a verbal beating everywhere because of it. He kept it up for a long time, then the posts stopped and a few years later their website switched from an AOL personal page under the name of kmetaverso to a dot com site and started touting large customers in its unique rather semi-literate fashion.
I recall him asking (on the Basslist or DIYloudspeakers list) for ideas for a business plan so they could get an SBA loan. I told him to market their speakers to little boutique mall shop chains like the Gap, etc as "Sound Spheres" that hang from the ceiling - a little searching might find that post. It appears they tout that sort of market as one of their customers on their website. Maybe I made them some money...
HK26147 said:
1) There is certainly diffraction and HOMs in folded horns. But at LFs these things are not as audible as they are at HF. The research indictates that 500 - 8 kHz is the only effective range of audibility and that this is a bell curve peaked at about 2 kHz. SO the LF folds would not be significant in terms of audibility. IF I were to use horns at LFs, I certainly would prefer not to fold them.
2) a little hard to follow the layout, but any sharp edge ahead of the source is going to be a significant diffraction effect and potentially audible.
HK26147 said:
Sometimes, the 'A' stands for Amateur - I've had plenty of disasters, but I guess that's the joys of R&D
.I was planning on starting a thread at some point, but was hoping to do so when I'd actually got something worthy (i.e. something that actually has a speaker driver in it)! I might get round to starting one with some pics of where I'm at - though I'll have to dig them out.
Ron E said:
Very interesting. I take it they're spraying it into a mold then?
Ron E said:
Yea... that is a touch dumb. I've seen a surprisingly high number of such enclosures over the years, though non high volume, probably due to the difficulties of manufacture. Trying to claims rights to a shape is unlikely to gain you much respect.
Ron E said:
I didn't want to mention it before, but yea, I had noticed the site wasn't the best designed I've seen.
Ron E said:I recall him asking (on the Basslist or DIYloudspeakers list) for ideas for a business plan so they could get an SBA loan. I told him to market their speakers to little boutique mall shop chains like the Gap, etc as "Sound Spheres" that hang from the ceiling - a little searching might find that post. It appears they tout that sort of market as one of their customers on their website. Maybe I made them some money...
That's the kind of market (well, maybe just wealthy individuals) I'm aiming for (should I ever actually produce anything worth selling). Hopefully he won't claim he has sole rights to sell such shapes
.You talk about two things, edge reflections/diffraction and GD is not the same thing.
Me and gedlee (and a couple of others) discussed this some time ago but the confusion seems to remain.
An edge reflection (such as gedlee has studied) will casue combfiltering and what you hear is an effect on frequency response and you can not make statements about audibility of GD on the basis of such a studie.
But groupdelay caused by for example the tuning of a loudspeakerbox or a crossover can sum perfectly flat and still have phase distortion.
And yes, groupdelay even at low frequencies is audible under certain cirumstances. There are a couple of studies that have found the limit to be about 1/f or perhaps 0.8/f. One exception is in the 100-250Hz range where I believe the sensitivity is a little higher on percussive sounds.
/Peter
Me and gedlee (and a couple of others) discussed this some time ago but the confusion seems to remain.
An edge reflection (such as gedlee has studied) will casue combfiltering and what you hear is an effect on frequency response and you can not make statements about audibility of GD on the basis of such a studie.
But groupdelay caused by for example the tuning of a loudspeakerbox or a crossover can sum perfectly flat and still have phase distortion.
And yes, groupdelay even at low frequencies is audible under certain cirumstances. There are a couple of studies that have found the limit to be about 1/f or perhaps 0.8/f. One exception is in the 100-250Hz range where I believe the sensitivity is a little higher on percussive sounds.
/Peter
Pan said:
Well, I must admit that my previous experience/understanding of diffraction issues was to do with the phantom source problem off sharp edges, leading to comb filtering.
Linking them to GD was something I hadn't come across before this thread, though because my understanding of GD isn't as good as I'd like it to be, I can't comment on the validity of this connection.
Pan said:
If you are saying that edge diffraction can occur with GD then you would need to show this mathematically not "the way I see it". Its either true and can be proven or its not.
To me edge diffraction is a multipath signal, which is know to be non-minimum phase and hence must have excess GD. And multipath signals will cause comb filtering, this is fundamental. Comb filtering is the frequency domain magnitude result and GD is the frequency domain phase result. In the time domain there are two impulses seperated in time. Thats the way I see it and it all fits together.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.